资源描述:
《Machine-Vision Detection for Rail-Steel’s Surface Flaws-4300.pdf》由会员上传分享,免费在线阅读,更多相关内容在学术论文-天天文库。
1、Proceedingsofthe7thWorldCongressonIntelligentControlandAutomationJune25-27,2008,Chongqing,ChinaMachine-VisionDetectionforRail-Steel’sSurfaceFlawsBasedonQuantumNeuralNetworkXueWangYikeTangandPingChengSchoolofElectric&InformationEngineeringCollegeofMechanic
2、EngineeringChongqingUniversityofScienceandTechnologyChongqingUniversityChongqing,400042,ChinaChongqing,400044,Chinasaltmaker@163.comtangyike@cqu.edu.cnAbstract-Conventionaldetectingmethodsbring2)Electriceddydetection:Whenaprobeinsideaelectricdisadvantages
3、oflow-efficiencyorhigh-falloutrateasforrail-coilwithsinuswavemotivationapproachingtoawaitingforsteel’ssurfaceflawsbecauseofitsnon-planarandsophisticatedinspectedsurface,inductionelectriceddywhichproducedbycontour.Ajournalmachinevisionapproachwaspresented,
4、inalternatingmagneticfieldaroundtheelectriccoil,wouldexistwhichimagingmethodandclassifieralgorithmareillustrated.onmetalsurface,andelectricproduceanti-magneticfluxatLinerCCDisadaptingtoimagingformovingrail-steel.Thethesamefrequencyandreversedirection.When
5、thepoleclassifierbasedonQuantumNeutralNetwork(QNN)algorithmcoulddealwiththosesimilarandhardlydifferentiatedROIofmeetsflaws,thealterationresistanceofelectriccoilindicatesflaws.Itdiscussedfeaturevectorparametersextractedfromthecategoryandmagnitudeoftheflaws
6、.differentspaces,moreover,QNN’smodel,multi-levelmotivation3)Infraredraydetection:FaradiccurrentisproducedonfunctionsbasedonSigmoidfunctionandtrainingalgorithmaresteelsurfacebyhigh-frequencyinductioncoil,andthefaradicexpatiatedindetail.Anexperimentaldevice
7、wasdevelopedanddepthwouldbelessthan1mm.Nearbytheflawarea,moretestresultsdemonstratethefeasibilityofthedetectionapproach.electricpowerisconsumedperunitlengthbecauseoffaradicIthasprovedtheeffectivenessandvalueofproposedmethodincurrent,solocaltemperatureonst
8、eelsurfaceiscalefactive.automaticdetectionforrail-steel’ssurfaceflaws.Thetemperaturedependsontheaveragedepthofflaws,thecoil’soperatingfrequencyandwidth,inputelectricpower,IndexTerms–Machine-vision,surfaceflaw,rail-s